test_imperative_gan.py 8.9 KB
Newer Older
X
Xin Pan 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
# Copyright (c) 2018 PaddlePaddle Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

import contextlib
import unittest
import numpy as np
import six
import sys

import paddle
import paddle.fluid as fluid
M
minqiyang 已提交
23
import paddle.fluid.core as core
X
Xin Pan 已提交
24
from paddle.fluid.optimizer import SGDOptimizer
25
from paddle.fluid import Conv2D, Pool2D, Linear
X
Xin Pan 已提交
26
from test_imperative_base import new_program_scope
L
lujun 已提交
27
from paddle.fluid.dygraph.base import to_variable
28
from paddle.fluid.framework import _test_eager_guard
X
Xin Pan 已提交
29 30


31
class Discriminator(fluid.Layer):
32 33 34 35
    def __init__(self):
        super(Discriminator, self).__init__()
        self._fc1 = Linear(1, 32, act='elu')
        self._fc2 = Linear(32, 1)
X
Xin Pan 已提交
36 37 38

    def forward(self, inputs):
        x = self._fc1(inputs)
39 40
        x = self._fc2(x)
        return x
X
Xin Pan 已提交
41 42


43
class Generator(fluid.Layer):
44 45 46 47 48
    def __init__(self):
        super(Generator, self).__init__()
        self._fc1 = Linear(2, 64, act='elu')
        self._fc2 = Linear(64, 64, act='elu')
        self._fc3 = Linear(64, 1)
X
Xin Pan 已提交
49 50 51 52

    def forward(self, inputs):
        x = self._fc1(inputs)
        x = self._fc2(x)
53 54
        x = self._fc3(x)
        return x
X
Xin Pan 已提交
55 56


L
lujun 已提交
57
class TestDygraphGAN(unittest.TestCase):
58
    def func_test_gan_float32(self):
X
Xin Pan 已提交
59
        seed = 90
C
cnn 已提交
60
        paddle.seed(1)
L
Leo Chen 已提交
61
        paddle.framework.random._manual_program_seed(1)
X
Xin Pan 已提交
62 63
        startup = fluid.Program()
        discriminate_p = fluid.Program()
X
Xin Pan 已提交
64 65
        generate_p = fluid.Program()

X
Xin Pan 已提交
66 67 68
        scope = fluid.core.Scope()
        with new_program_scope(
                main=discriminate_p, startup=startup, scope=scope):
69 70
            discriminator = Discriminator()
            generator = Generator()
X
Xin Pan 已提交
71 72 73 74 75 76 77 78 79

            img = fluid.layers.data(
                name="img", shape=[2, 1], append_batch_size=False)
            noise = fluid.layers.data(
                name="noise", shape=[2, 2], append_batch_size=False)

            d_real = discriminator(img)
            d_loss_real = fluid.layers.reduce_mean(
                fluid.layers.sigmoid_cross_entropy_with_logits(
X
Xin Pan 已提交
80 81 82
                    x=d_real,
                    label=fluid.layers.fill_constant(
                        shape=[2, 1], dtype='float32', value=1.0)))
X
Xin Pan 已提交
83 84 85 86

            d_fake = discriminator(generator(noise))
            d_loss_fake = fluid.layers.reduce_mean(
                fluid.layers.sigmoid_cross_entropy_with_logits(
X
Xin Pan 已提交
87 88 89
                    x=d_fake,
                    label=fluid.layers.fill_constant(
                        shape=[2, 1], dtype='float32', value=0.0)))
X
Xin Pan 已提交
90 91 92 93 94 95 96

            d_loss = d_loss_real + d_loss_fake

            sgd = SGDOptimizer(learning_rate=1e-3)
            sgd.minimize(d_loss)

        with new_program_scope(main=generate_p, startup=startup, scope=scope):
97 98
            discriminator = Discriminator()
            generator = Generator()
X
Xin Pan 已提交
99 100 101 102 103 104 105

            noise = fluid.layers.data(
                name="noise", shape=[2, 2], append_batch_size=False)

            d_fake = discriminator(generator(noise))
            g_loss = fluid.layers.reduce_mean(
                fluid.layers.sigmoid_cross_entropy_with_logits(
X
Xin Pan 已提交
106 107 108
                    x=d_fake,
                    label=fluid.layers.fill_constant(
                        shape=[2, 1], dtype='float32', value=1.0)))
X
Xin Pan 已提交
109 110 111 112

            sgd = SGDOptimizer(learning_rate=1e-3)
            sgd.minimize(g_loss)

M
minqiyang 已提交
113 114
        exe = fluid.Executor(fluid.CPUPlace() if not core.is_compiled_with_cuda(
        ) else fluid.CUDAPlace(0))
X
Xin Pan 已提交
115
        static_params = dict()
X
Xin Pan 已提交
116 117 118 119
        with fluid.scope_guard(scope):
            img = np.ones([2, 1], np.float32)
            noise = np.ones([2, 2], np.float32)
            exe.run(startup)
X
Xin Pan 已提交
120 121 122 123 124 125 126 127 128
            static_d_loss = exe.run(discriminate_p,
                                    feed={'img': img,
                                          'noise': noise},
                                    fetch_list=[d_loss])[0]
            static_g_loss = exe.run(generate_p,
                                    feed={'noise': noise},
                                    fetch_list=[g_loss])[0]

            # generate_p contains all parameters needed.
X
Xin Pan 已提交
129
            for param in generate_p.global_block().all_parameters():
X
Xin Pan 已提交
130 131 132 133
                static_params[param.name] = np.array(
                    scope.find_var(param.name).get_tensor())

        dy_params = dict()
L
lujun 已提交
134
        with fluid.dygraph.guard():
C
cnn 已提交
135
            paddle.seed(1)
L
Leo Chen 已提交
136
            paddle.framework.random._manual_program_seed(1)
X
Xin Pan 已提交
137

138 139 140 141 142 143
            discriminator = Discriminator()
            generator = Generator()
            sgd = SGDOptimizer(
                learning_rate=1e-3,
                parameter_list=(
                    discriminator.parameters() + generator.parameters()))
X
Xin Pan 已提交
144 145 146 147 148 149 150 151 152 153 154 155 156

            d_real = discriminator(to_variable(np.ones([2, 1], np.float32)))
            d_loss_real = fluid.layers.reduce_mean(
                fluid.layers.sigmoid_cross_entropy_with_logits(
                    x=d_real, label=to_variable(np.ones([2, 1], np.float32))))

            d_fake = discriminator(
                generator(to_variable(np.ones([2, 2], np.float32))))
            d_loss_fake = fluid.layers.reduce_mean(
                fluid.layers.sigmoid_cross_entropy_with_logits(
                    x=d_fake, label=to_variable(np.zeros([2, 1], np.float32))))

            d_loss = d_loss_real + d_loss_fake
L
lujun 已提交
157
            d_loss.backward()
X
Xin Pan 已提交
158
            sgd.minimize(d_loss)
X
Xin Pan 已提交
159 160
            discriminator.clear_gradients()
            generator.clear_gradients()
X
Xin Pan 已提交
161

X
Xin Pan 已提交
162 163 164 165 166
            d_fake = discriminator(
                generator(to_variable(np.ones([2, 2], np.float32))))
            g_loss = fluid.layers.reduce_mean(
                fluid.layers.sigmoid_cross_entropy_with_logits(
                    x=d_fake, label=to_variable(np.ones([2, 1], np.float32))))
L
lujun 已提交
167
            g_loss.backward()
X
Xin Pan 已提交
168 169
            sgd.minimize(g_loss)
            for p in discriminator.parameters():
170
                dy_params[p.name] = p.numpy()
X
Xin Pan 已提交
171
            for p in generator.parameters():
172
                dy_params[p.name] = p.numpy()
X
Xin Pan 已提交
173

174 175
            dy_g_loss = g_loss.numpy()
            dy_d_loss = d_loss.numpy()
X
Xin Pan 已提交
176

177 178
        dy_params2 = dict()
        with fluid.dygraph.guard():
179
            fluid.set_flags({'FLAGS_sort_sum_gradient': True})
C
cnn 已提交
180
            paddle.seed(1)
L
Leo Chen 已提交
181
            paddle.framework.random._manual_program_seed(1)
182 183 184 185 186 187
            discriminator2 = Discriminator()
            generator2 = Generator()
            sgd2 = SGDOptimizer(
                learning_rate=1e-3,
                parameter_list=(
                    discriminator2.parameters() + generator2.parameters()))
188 189 190 191 192 193 194 195 196 197 198 199 200

            d_real2 = discriminator2(to_variable(np.ones([2, 1], np.float32)))
            d_loss_real2 = fluid.layers.reduce_mean(
                fluid.layers.sigmoid_cross_entropy_with_logits(
                    x=d_real2, label=to_variable(np.ones([2, 1], np.float32))))

            d_fake2 = discriminator2(
                generator2(to_variable(np.ones([2, 2], np.float32))))
            d_loss_fake2 = fluid.layers.reduce_mean(
                fluid.layers.sigmoid_cross_entropy_with_logits(
                    x=d_fake2, label=to_variable(np.zeros([2, 1], np.float32))))

            d_loss2 = d_loss_real2 + d_loss_fake2
201
            d_loss2.backward()
202 203 204 205 206 207 208 209 210
            sgd2.minimize(d_loss2)
            discriminator2.clear_gradients()
            generator2.clear_gradients()

            d_fake2 = discriminator2(
                generator2(to_variable(np.ones([2, 2], np.float32))))
            g_loss2 = fluid.layers.reduce_mean(
                fluid.layers.sigmoid_cross_entropy_with_logits(
                    x=d_fake2, label=to_variable(np.ones([2, 1], np.float32))))
211
            g_loss2.backward()
212 213 214 215 216 217 218 219 220
            sgd2.minimize(g_loss2)
            for p in discriminator2.parameters():
                dy_params2[p.name] = p.numpy()
            for p in generator.parameters():
                dy_params2[p.name] = p.numpy()

            dy_g_loss2 = g_loss2.numpy()
            dy_d_loss2 = d_loss2.numpy()

X
Xin Pan 已提交
221 222 223 224
        self.assertEqual(dy_g_loss, static_g_loss)
        self.assertEqual(dy_d_loss, static_d_loss)
        for k, v in six.iteritems(dy_params):
            self.assertTrue(np.allclose(v, static_params[k]))
X
Xin Pan 已提交
225

226 227 228 229 230
        self.assertEqual(dy_g_loss2, static_g_loss)
        self.assertEqual(dy_d_loss2, static_d_loss)
        for k, v in six.iteritems(dy_params2):
            self.assertTrue(np.allclose(v, static_params[k]))

231 232 233 234 235
    def test_gan_float32(self):
        with _test_eager_guard():
            self.func_test_gan_float32()
        self.func_test_gan_float32()

X
Xin Pan 已提交
236 237 238

if __name__ == '__main__':
    unittest.main()